Over the last few years, the United States Department of Housing and Urban Development has sought ways to learn patterns of service utilization across homeless programs, while guaranteeing the privacy of those clients who visit domestic violence homeless shelters. This talk reports on the surprises of what didn’t work and why, and then introduces PrivaMix, a real-time, multi-party computation invented as a solution to this problem. A contribution is the use of a one-way function that has a commutative property. A PrivaMix function assigns made-up identifiers such that the identifiers for the same client at different shelters are distinct, yet the identifiers can be computationally “mixed” using the PrivaMix Protocol to securely link records belonging to the same client. Real-world experiments in Des Moines, Iowa showed that PrivaMix worked flawlessly. PrivaMix is now expanding into real-world healthcare applications to link de-identified patient records across institutions. As a supplement to this talk, Sweeney will discuss the seven-year history of the Data Privacy Lab at Carnegie Mellon by briefly mentioning a summary of her work performed and lessons learned for a new future.
Latanya Sweeney creates technology that weaves with policy to resolve real-world technology/privacy clashes. She is a distinguished career professor of computer science, technology and policy at Carnegie Mellon University, and founder and director of the Data Privacy Lab. She has received numerous awards, and testified before federal and international government bodies, and in 2009, was appointed to the Federal Health Information Technology Policy Committee.
